Android Profitability, Revisited
by Mark W. Hibben
That investors were mostly dismayed by Google’s Q1 2011 financial results and conference call could be seen in the vicious beating the stock took the day after. The news was not all bad, however, as a careful examination of Google’s financials will show. The problems evident in the Q1 results are almost certainly due to their Android strategy. Google execs continue to be very tight lipped about Android in general, and this isn’t helping analysts’ perceptions. From the outside, it appears that Google’s costs are out of control, a result of apparently undisciplined expansion. I suspect this isn’t the case, and that cost growth is the result of a very deliberate strategy to capture and hold market share for Android. The War for Mobile Internet Supremacy (WMIS) is turning out to be an expensive undertaking for Google, but they know that they must win, have known, probably longer than anyone else. Furthermore, the ad based revenue model that has worked well for Google’s core business of search does appear to be able to sustain development and maintenance of a competitive Mobile OS, and may even lead to Google’s eventual victory in the WMIS.
Android Global Population Update
Before I pick apart Google’s financials, let’s briefly update the Android population model. As the reader may recall from the previous article Will Android be Profitable? My financial model for Google’s revenues and costs is based on an estimate of the population of devices that can host Google ads and sites. As the population of devices grows, both mobile and desktop, so grows Google’s business, both revenue and costs. It’s a simplified model, but it works very well as we shall see for revenue and for Google’s major cost, called revenue cost, which includes so called traffic acquisition costs. Therefore it’s very important to have a handle on Android, desktop and mobile device populations in general. Let’s start with Android.
At the conference call, Google execs for the first time confirmed that Android activation rate is now “greater than” 350000 units/day, a significant milestone. I went back to my Android population model all set to tweak it to fit the new data. It is, after all, an empirical curve fit. I found I didn’t need to. The announced activation rate is already a good fit to the curve, and all I had to do was add data points for the Q1 2011 results. The Android activation rate is a curve fit based on Google’s own data on activation rate announced during the various quarterly conference calls. The curve fit isn’t a smooth curve because I only calculate activation rate on a quarterly basis. The android population is estimated by numerically integrating the activation rate curve. This integration amounts to simply summing the total android devices in existence with the newly activated devices.
As I previously pointed out, Gartner’s Mobile OS population data as of the end of 2010 had overstated the market share of Android OS, especially relative to Apple iOS. It was to arrive at a more reliable estimate of the Android OS population that I adopted this “integration” approach to calculating Android OS population. According to this model, Android OS population now tops out at a little more than 113 million devices worldwide.
Has Android finally beaten Apple in mobile OS market share? No, not yet. According to Apple, as of their January 18, 2011 quarterly financial conference call, Apple had shipped 167 million iOS devices. This includes iPhones, iPads, and iPod Touch devices. However, if we exclude iPod Touch, which might be a more fair comparison, then the total deliveries of iPhones and iPads as of the end of 2010 stood at 104.7 million. Apple is more forthcoming on device deliveries, reporting specific numbers for iPhones, iPads, Macs and iPods every quarter, so I just use the Apple data. Assuming a modest 10% replacement rate for iPhones and zero for iPads, then the total population of iPhones and iPads at the end of 2010 stood at 95.8 million. Apple is due to report their Q2 results this week, and I expect that the iPhone and iPad population will have grown by at least 20 million, keeping Apple slightly ahead of Android in this tally, but we’ll see.
Google Revenue and Cost Model
As a mass media advertising business, Google’s revenues are very much dependent on the total number of devices that can host Google ads or that can be used to visit Google sites. Here, the revenue model is very simple and agrees relatively well with Google’s actuals:
Total Revenue ($) = Device Revenue ($) * Number of Devices
Device Revenue is the average revenue per device per quarter and is assumed to be fixed over time. I derive an initial estimate of device revenue from Google’s revenue actuals divided by the total number of desktop and relevant mobile devices. Currently, I only include iOS and Android devices in the mobile category, under the assumption that RIM, Symbian and Windows Mobile devices are infrequently used for Web browsing. Currently I estimate a Device Revenue of $4.8 /device-quarter. Probably there is some variation in Device Revenue according to device type, but, lacking any additional information from Google, I have no way to estimate that difference. Device number is the dominant term anyway, since there are an estimated 1.47 B desktops worldwide, while there are only about 290 M iOS and Android devices. Shown below is a comparison of the revenue model results compared to Google’s actuals.
Note that Android, based on population, makes a relatively small contribution to Google revenues, with the gap between total and desktop revenues made up by iOS devices. The short term fluctuations in the revenue actuals are to be expected, and amount to “noise”, i.e., statistical fluctuations in the real data. Also note that here iOS devices includes iPod touch, which is as usable for Web browsing as iPhone and is therefore included.
Google’s Revenue Cost, which includes Traffic Acquisition Cost, is directly proportional to revenue, and therefore revenue costs for desktop and mobile devices should also be proportional to the respective revenues.
It’s not clear yet whether the recent decline in Revenue Cost represents a trend or is merely “noise”. If it is a trend, then it’s good news for Google investors, but it’s too early to tell. The general trend is that revenue and revenue cost rise proportionately to the increase in the worldwide population of mobile and desktop computers. This is good for Google and Google’s investors, since it provides guaranteed revenue growth.
One wouldn’t necessarily expect Google’s other major cost categories to also grow proportionately with device population. After all, there’s no direct linkage between G&A, Sales and Marketing, and Research and Development to device population, and Google management are more or less free to spend as they see fit. However, if Google management are “living within their means”, then the long term trend in these spending categories should correlate with revenue, and therefore with device population growth as well.
We can test this hypothesis by applying the population growth model to these other costs, and seeing how well the actuals agree. For G&A, the model agreement is pretty good:
Sales and Marketing costs diverge significantly from the model for the previous and current quarters:
Similarly, R&D cost diverges from the model for the current quarter:
Another way to look at the change in Google’s cost structure is a comparison of pie charts for the previous and current reported quarters:
While G&A has held steady, and Revenue Cost has actually decreased (relatively), both Marketing and R&D have increased.
The deviations from the device population models for Sales and Marketing and Research and Development costs certainly convey that Google management are making some “investments” that they expect to pay off downstream. Unfortunately, neither the prepared statements nor answers to analysts’ questions at the conference call shed much light on what these “investments” might be. Normally Sales and Marketing would capture costs related to marketing Google’s principal product, advertising, to its advertiser customer base, but there was little in the conference call to suggest a big push in this regard. However, Sales and Marketing could also capture marketing costs associated with promotion of any of Google’s other products including Chrome browser, Chrome OS, Android, and Google search.
It’s possible that support (both technical and marketing) to Google’s Android partners (such as Motorola and HTC) could even be captured as a “marketing” cost, and it’s certainly not apparent where else such support costs would be booked. My strong suspicion is that Android is responsible for a significant portion of the Sales and Marketing cost in excess of the model, perhaps as much as half of the excess amount or about US$ 75 M. Clearly Chrome OS and Android OS must be responsible for most, if not all, of the R&D cost in excess of the model, about US$ 100 M. If we assume that this is split evenly between Chrome OS and Android OS, then this comes to about US$ 50 M each.
Was Android Profitable?
Certainly, according to my population-based revenue and cost model, Android, treated as a stand-alone business, was profitable in Q1, to the tune of US$ 188M as the chart shows:
I estimate Android-related revenue to have been US$ 540 M in Q1, while Android-related costs are estimated to be US$ 352 M. More importantly, even if we make a worst case assumption about Google’s excess Sales and Marketing and R&D costs for Android, this amounts to a subtraction from Android Income of another US$ 125 M, and Android still squeaks by with a profit, albeit a measly US$ 63 M. The unusual advertising based revenue model for Android is creaking under the strain but appears to be holding up. In a sense, iOS partially subsidizes Android, by providing another revenue stream on which Google may draw.
All this bodes very well for the long term prospects of Android. As an advertising supported “free” mobile OS, it will always be less expensive initially to the consumer than iOS or Windows Phone. Furthermore, Google is free to build in “features” that enhance revenue per Android device compared to other devices, as well as reduce revenue cost per Android device, thereby making Android more profitable relative to other devices. I expect that if current trends continue, Android population, as well as Android Income from Operations, will surpass that of iOS some time in the not too distant future.